Digital discrimination in the sharing economy: what’s mine is not yours : evidence from an online experiment

(1)

Norwegian School of Economics

Digital Discrimination in the Sharing Economy: What’s Mine is Not Yours

Evidence From an Online Experiment

Master Thesis - Marketing and Brand Management

Sigrid Klemsdal and Pauline Sundt

Supervised by Siv Skard

This thesis was written as a part of the Master of Science in Economics and Business Administration at NHH. Please note that neither the institution nor the examiners are responsible — through the approval of this thesis — for the theories and methods used, or results and conclusions drawn in this work.

Bergen, Spring 2017

(2)

Abstract

Although the issue of racial discrimination has been studied extensively throughout the past decades, its appearance in the disruptive digital market called the sharing economy is a rather unexplored field of research. To address this issue, we study how consumer outcomes in an Airbnb- setting are influenced by a host’s Arab (out-group) ethnicity compared to a Norwegian (in-group) ethnicity, and the underlying mechanisms and boundary conditions involved. Further, we examine the effect of introducing in-group characteristics (an in-group symbol) to the out-group host, to potentially eliminate discrimination.

Conducting an online survey experiment, we find negative main effects of the Arab host’s ethnicity (vs. a Norwegian ethnicity) on some consumer outcomes. Yet, additional significant effects arise when accounting for individual differences in political orientation and intergroup threat. More specifically, being right-oriented (left-oriented) and perceiving Muslims as being of high (low) threat towards Western culture strengthens (reduces) out-group discrimination. Trustworthiness appears as the most important explanatory mechanism of the effects. In addition, the in-group symbol seems to eliminate discrimination in general, however not for all levels of political orientation and intergroup threat.

The uncovering of racial discrimination on Airbnb has impactful consequences for decision makers in the sharing economy. As the issue of digital discrimination is likely to persist as we move into the future, it is crucial for such platforms to focus on how this can be minimized. A key challenge is to properly facilitate trust among all users, no matter their origin - to create communities where trulyeveryone can belong.

Key words: Sharing Economy, Racial Discrimination, Social Identity, Self-other Overlap, Self- brand Connection, Trustworthiness, Perceived Risk, Political Orientation, Intergroup Threat, Con- sumer Outcomes, Marketing.

(3)

Acknowledgements

This thesis is part of our Master’s degrees in Marketing and Brand Management (MBM) at the Norwegian School of Economics (NHH). Five years of studying Economics and Business Admin- istration have now passed, and this thesis symbolizes the end of our education. Even though the research process has certainly been demanding, we have appreciated the challenge, and acquired useful skills to bring with us into the future.

Studying a topic as important as racial discrimination has truly been both rewarding and eye- opening. We feel honored to have been part of a research project emphasizing the importance of equality and belonging for all citizens of our world.

We would like to thank Adjunct Associate Professor at NHH, Siv Skard, for supervising our thesis.

Her excellent guidance and devotion throughout the whole process has been valuable, inspiring and very much appreciated. Furthermore, we would like to thank PhD Candidates Hallgeir Sj˚astad and Katrine Nødtvedt for helpful input when preparing the experiment. Finally, we would like to thank The Center for Service Innovation (CSI) at NHH for funding the research project.

Bergen, June 2017

Sigrid Klemsdal and Pauline Sundt

(4)

List of Figures

1.1 Proposed Model of Effects . . . 12

3.1 Proposed Model of Effects With Hypotheses . . . 35

5.1 Simple Moderation Model (Hayes 2013) . . . 51

5.2 Simple Mediation Model (Hayes 2013) . . . 52

5.3 Parallel Multiple Mediator Model Withj Mediators (Preacher and Hayes 2008) . . 53

5.4 Conditional Process Model (Hayes 2013) . . . 54

5.5 Parallell Mediation Model: Willingness to Switch Apartments . . . 64

5.6 Parallell Mediation Model: Willingness to Rent . . . 65

5.7 Overview of Proposed Model of Effects . . . 66

6.1 Simple Moderation Model: Trustworthiness and Political Orientation . . . 73

A.1 Fictive Apartment Ad for White Host . . . 105

A.2 Fictive Apartment Ad for Arab Hosts . . . 106

A.3 Fictive Host Profile for White Host . . . 107

A.4 Fictive Host Profile for Arab Host . . . 107

A.5 Fictive Host Profile for Arab Host with In-group Symbol . . . 108

F.1 Parallel Mediation Model: Liking of Apartment . . . 123

F.2 Parallel Mediation Model: Attributes of Apartment . . . 123

F.3 Parallel Mediation Model: Willingness to Pay . . . 124

(8)

List of Tables

5.1 Mean Scores, Dependent Variables . . . 57

5.2 Areas of Significance for Moderators . . . 60

C.1 Factor Loadings and Cronbach’s Alpha, Attributes of Apartment . . . 115

C.2 Factor Loadings and Cronbach’s Alpha, Trustworthiness . . . 115

D.1 Descriptive Statistics, Scenario 1 . . . 116

D.2 Descriptive Statistics, Scenario 2 . . . 116

D.3 Descriptive Statistics, Moderators . . . 116

D.4 Levene’s Test of Equality of Variance . . . 117

E.1 Planned Comparison, White vs. Arab . . . 118

E.2 Descriptives, Attractiveness of Apartment . . . 118

E.3 Descriptives, Willingness to Switch Apartments . . . 118

E.4 Planned Comparison, White vs. Arab Symbol . . . 119

F.1 Mediation Effects of Host Ethnicity on Liking of Apartment . . . 120

F.2 Mediation Effects of Host Ethnicity on Attributes of Apartment . . . 120

F.3 Mediation Effects of Host Ethnicity on Attractiveness of Apartment . . . 121

F.4 Mediation Effects of Host Ethnicity on Willingness to Rent . . . 121

F.5 Mediation Effects of Host Ethnicity on Willingness to Pay . . . 122

F.6 Mediation Effects of Host Ethnicity on Willingness to Switch Apartments . . . 122

G.1 Moderated Mediation: Political Orientation on Liking of Apartment . . . 125

G.2 Moderated Mediation: Political Orientation on Attributes of Apartment . . . 126

G.3 Moderated Mediation: Political Orientation on Attractiveness of Apartment . . . . 126

G.4 Moderated Mediation: Political Orientation on Willingness to Rent . . . 127

G.5 Moderated Mediation: Political Orientation on Willingness to Pay . . . 127

G.6 Moderated Mediation: Political Orientation on Willingness to Switch Apartments . 128 G.7 Moderated Mediation: Intergroup Threat on Liking of Apartment . . . 128

(9)

G.8 Moderated Mediation: Intergroup Threat on Attributes of Apartment . . . 129

G.9 Moderated Mediation: Intergroup Threat on Attractiveness of Apartment . . . 129

G.10 Moderated Mediation: Intergroup Threat on Willingness to Rent . . . 130

G.11 Moderated Mediation: Intergroup Threat on Willingness to Pay . . . 130

G.12 Moderated Mediation: Intergroup Threat on Willingness to Switch Apartments . . . 131

(10)

Chapter 1: Introduction

1.1 Background

Our world is currently experiencing a drastic and highly impactful wave of digitalization. 8 billion devices are now connected to the internet, and this number is predicted to reach as much as 1 trillion in 2030 (WEF 2017). As a result, consumers are engaging in new ways and demonstrating completely new behaviors (Qvartz 2016). We no longer take part in just the physical world - the virtual world has become an equally essential aspect of our lives. In recent years, this rapid digitalization has triggered the rise of several disruptive markets. Built solely on the basis of technology, these businesses represent the new digital economy (PwC 2011).

A particularly flourishing part of the digital market is the sharing economy (PwC 2015). Over the last decade, a number of new businesses using the traditional concept of sharing as a central element of their business models have emerged. The sharing economy, also called collaborative consumption (Botsman and Rogers 2010) has been called an “idea that will change the world” (Time 2011).

Consumers all around the world are taking part in this new marketplace, destined to make our lives easier - and more efficient (Leong 2015). By sharing everything from housing (Airbnb) to cars (Zipcar) to screwdrivers (Taskrabbit), this economy enables access to underutilized resources, profiting both owners, users, and Mother Earth. Valued as a EUR 28 billion industry in Europe in 2016, Europe’s five most prominent sharing economy sectors are predicted to deliver EUR 570 billion by 2025 (PwC 2016). Evidently, there is no doubt that this groundbreaking economy is here to stay.

However, in a peer-to-peer economy based purely on private individuals deciding themselves who to share their own resources with, there is a risk of certain biases and prejudice unfolding (Leong 2015).

Especially one age-old problem still remains in most societies: the issue of racial discrimination.

Sadly, as witnessed across the globe, racial discrimination is not only a concept for the history books, but also very much relevant in our society today. Extensive research within housing,

(11)

employment, mortgage lending and a wide range of other social areas has exposed how racial discrimination is an evident issue the modern world (Van Bavel and W. A. Cunningham 2009). As a result, the issue of racial discrimination plays a critical part in shaping our society, by continuing to affect the allocation of contemporary opportunities (Pager and Shepherd 2008).

Yet, with the sharing economy’s innovative way of virtually connecting people, could this disruptive market lead to a final eradication of such prejudicial behavior? Many have considered the various new businesses of the sharing economy to potentially pose as a cure for racial discrimination, by filtering out the racial signifiers one would encounter in more traditional marketplaces (Leong 2015).

Instead of meeting each other in person, we communicate through digital channels, providing only the most essential information related to the transaction. However, recent incidents in the sharing economy raise questions as to whether such online platforms really reduce discrimination - or if they actually worsen it.

In March of 2015, a twenty-five-year-old Black man from Virginia named Gregory Selden was planning a trip to Philadelphia (Vara 2017). Eager to try out the Airbnb service for the first time, he set up his profile with a picture and some basic information about himself, and requested a place he liked. Although the apartment was posted as vacant for the relevant dates, the owner quickly messaged him back saying it was unavailable. As Selden immediately thought there was something strange about the exchange, he decided to create fake Airbnb profiles for two White people named ”Jessie” and ”Todd”, and sent in requests for the same property and the same dates.

In just moments after, Selden’s suspicions seemed to be proven right: Jessie and Todd were both welcome.

Appalled by how the owner seemed to refuse to accommodate him simply based on his race, Selden turned to social media to share his experience (Vara 2017). In turn, this inspired many others to come forward with their own stories of similar instances. By using the tag ”AirbnbWhileBlack”, the personal stories were spread quickly throughout the internet. However, it was difficult to know whether these occurrences were only accidental, or if users were actually being discriminated solely based on their race. Yet in 2016, this phenomenon was picked up by three researchers - and the results were just as alarming.

(12)

In an experiment conducted at Harvard by creating fictional profiles on Airbnb, it was revealed how applications from guests with distinctively African-American names were 16 percent less likely to be accepted relative to identical guests with distinctively White names (Edelman, Luca, and Svirsky 2017). This article was, to the best of our knowledge, the first study to expose how racial bias influences consumer decision making within the sharing economy. As a market that has skyrocketed the last couple of years, and is predicted to continue growing far into the future, the issue of racial discrimination is unarguably an important area to investigate further. This sparks the question: are we moving into an era of digital discrimination?

1.2 Purpose

The purpose of this study is to investigate the newly identified phenomenon of racial discrimination within the sharing economy. More specifically, we will investigate how the ethnicity of a host on Airbnb influences consumers’ attitudes and behavioral intentions, both when it comes to renting the host’s apartment and switching their own apartment with the host’s. We will look into potential processing mechanisms in this relationship, in order to learn more about the underlying psychological causes of discrimination. Also, we will study the boundary conditions involved in this overall process.

In addition, we will examine a potential strategy of eliminating bias towards the host, by introducing an ”in-group symbol” meant to induce a common identity. As will be discussed during the literature review, there are certain ways of manipulating the perceived relation to another person, which can ultimately lead to more favorable outcomes for the person in question.

Based on this, we will seek to answer the following research questions:

RQ₁: Will Airbnb users discriminate hosts/guests based on ethnicity?

RQ₂: What are the psychological mechanisms explaining ethnic discrimination on Airbnb?

RQ₃: What are the boundary conditions for these effects?

RQ₄: Can in-group symbols eliminate ethnic discrimination?

(13)

In order to clarify the overall structure of the literature review in the following chapter, the proposed model is presented below. This is a visual representation of our study, and illustrates the variables and relationships we will explore.

Figure 1.1: Proposed Model of Effects

As shown in Figure 1.1, the effect of discrimination will be examined both directly and indirectly, by investigating the psychological processes behind this relationship. The concepts we will explore are self-other overlap, self-object connection, trustworthiness and perceived risk. Classified as theories of congruence, self-other overlap and self-object connection (based on self-brand connection) encompass how we perceive and evaluate the world around us - with the aim of maintaining internal consistency (Abelson et al. 1968) and enhancing our self-concept (Graeff 1996). In addition, the theories of trust and perceived risk are both proved to be important within the area of consumer behavior, particularly in digital markets (Harridge-March 2006).

Furthermore, potential moderators of both the direct and indirect mechanisms will be identified.

More specifically, we will investigate whether individual differences in political orientation and perceived intergroup threat related to a foreign ethnic group may reveal differences in outcomes.

This will provide more specific information regarding how individual traits influence discriminatory thoughts and behavior.

(14)

As will be discussed in the literature review, all of the abovementioned theoretical constructs often appear within the field of consumer psychology, including in research on racial prejudice and discrimination. Overall, by learning more about the underlying factors explaining racial discrimination in this market, we can provide recommendations for potential improvements on sharing platforms, with the ultimate goal of eliminating it.

1.3 Structure

This master thesis starts with a review of the literature related to its three main topics: the sharing economy, racial discrimination and relevant research within the field of consumer psychology. As the merging of the research on the sharing economy and racial discrimination is relatively novel, the topics will first be introduced in separate parts, and then brought together. Furthermore, the theoretical basis for the psychological processes explaining discrimination will be presented. This is followed by a discussion of how this study can contribute to the research field. After this, the proposed model and the development of hypotheses will be described in more detail. Then, the methodology and findings of the study are presented. Finally, we will conclude by discussing the implications of our findings, the limitations of our study and suggestions for further research.

(15)

Chapter 2: Literature Review

2.1 Literature Search Process

This study involves the merging of three broad topics: 1) the sharing economy, 2) racial discrimination and 3) the psychological mechanisms behind discrimination. The literature search process involved extensive research within these areas, with the aim of achieving a comprehensive basis for our research. The database Business Source Complete was used as the starting point for all three topics, both separately and combined. In addition, the databases PsycInfo and PsycArticles were used as supplements when researching topics related to psychology. Google Scholar was also utilized when neither of the abovementioned databases provided sufficient numbers or types of articles on a certain topic.

2.2 Main Focus of Previous Research

2.2.1 The Sharing Economy

Sharing can be seen as one of the most basic forms of human economic behavior and has existed as a form of exchange in the society for hundreds of years (Prince 1975). In recent years, however, the Internet and other associated technologies have made sharing possible on a substantially larger scale (B. Cohen and Kietzmann 2014). Sharing platforms act as an intermediate empowering individuals to distribute, share and reuse, and provide transparency and convenience (Parsons 2014; Kathan, Matzler, and Veider 2016). The phenomenon as we know it today was first described by Botsman and Rogers (2010) in their bookWhat’s Mine is Yours, and has since gained widespread popularity among scholars, platforms and users. Due to its newness and diversity in the services offered, the sharing economy lacks a uniform definition. We will, however, use the commonly cited description of collaborative consumption: “an economic model based on sharing, swapping, trading, or renting

(16)

products and services, enabling access over ownership” (Botsman 2013, p. 1).

Airbnb, the “canonical example of the sharing economy” (Edelman, Luca, and Svirsky 2017), is an online marketplace for short-time housing rentals. Airbnb facilitates transactions between hosts and guests by enabling advertising of apartments, communication and handling payment, among other things. Hosts may decide whether to accept or reject a guest after seeing the person’s name, and often a picture. Likewise, potential guests may utilize the information available about the host when selecting an apartment.

Previous research has unveiled the key reasons as to why people choose to take part in sharing economy platforms. Participation is motivated by factors such as sustainability, enjoyment of the activity as well as economic gains (Hamari, Sj¨oklint, and Ukkonen 2016). In addition, the sense of community belonging, social interactions, and altruistic enjoyment of helping others are part of the socially-related drivers highly relevant for participating in this unique marketplace (Teubner, Hawlitschek, and Gimpel 2016). Members expect to derive happiness and positive emotions from the socialization and community bonding arising from sharing (Hellwig et al. 2015). This is in line with Airbnb’s claims on their website, to create an experience in which each guest will be a part of

“a trusted community marketplace” (Airbnb 2017a). As mentioned in an industry report on the sharing economy:

”Today, the value of a [sharing economy] brand is often linked to the social connections it fosters.

. . . By providing consumers with ease of use and confidence in decision-making, a [sharing economy] company moves beyond a purely transaction-based relationship to become a platform for an experience – one that feels more like friendship” (PwC 2015, p. 15).

However, how truly inclusive and fostering of friendly social interactions is the sharing economy?

An upcoming section seeks to answer this question, by focusing on a potential issue of racial discrimination. But first, in order to achieve a general overview of the racial discrimination literature, this will be presented in the following part.

(17)

2.2.2 Previous Research on Racial Discrimination

The issue of racial discrimination has been investigated in a vast range of social and economic areas. In order to provide a summary of the research relevant to this thesis, this will now briefly be discussed.

Racial Discrimination in Marketplaces and Marketing Communications

Racial discrimination has undoubtedly played a crucial part in many historical events through both ancient and modern times (Van Bavel and W. A. Cunningham 2009). With the purpose of establishing a universal description of the issue, The International Convention on the Elimination of All Forms of Racial Discrimination defines racial discrimination as “any distinction, exclusion, restriction or preference based on race, colour, descent or national or ethnic origin which has the purpose or effect of nullifying or impairing the recognition, enjoyment or exercise, on an equal footing, of human rights and fundamental freedoms in the political, economic, social, cultural or any other field of public life” (UN 1965, p. 2). As emphasized by the definition, racial discrimination can come in many shapes and forms, and can be related to nearly any type of context.

Although progress has been made in most societies, evidence collected over the past decades portray how racial minorities still face considerable discriminatory behavior in a vast array of marketplaces. Such market transactions could be renting an apartment (Hanson, Hawley, and A. Taylor 2011), getting a job (Bye et al. 2014), being accepted for a bank loan (Harkness 2016), or shopping, both in a store (Bennett, Hill, and Daddario 2015) and online (Nunley, Owens, and Howard 2015). In addition, racial discrimination has been established within pricing (Ayres and Siegelman 1995; A. T. King and Mieszkowski 1973). These inequalities can be witnessed across the globe. Within the area of housing alone, a growing breadth of research shows how ethnicity- based discrimination is apparent in Western societies that attract migrants, including Norway (for a review, see Hanson and Hawley 2014). Overall, the persistent racial inequalities in most societies have renewed interest in the problem of racial discrimination, as it continues to skew thereal social and economic opportunities people are facing (Pager and Shepherd 2008).

(18)

An area that particularly reflects how certain races are undervalued in a society is the field of marketing communications. As stated by D. Cohen (1970, p. 3): ”it has become almost axiomatic [unquestionable] to claim that advertising reflects the culture and the society in which it exists”.

Marketers often design brand positioning strategies by focusing on their majority consumers (Ben- nett, Hill, and Oleksiuk 2013), in order to best generate positive beliefs and attitudes, purchasing behavior and eventually brand loyalty (Pullig, Netemeyer, and Biswas 2006). As a result, minority consumers are often excluded from advertisements altogether (Puntoni, Vanhamme, and Visscher 2011).

Furthermore, when depicted in marketing communications, minorities are often portrayed by using stereotypes -”consensually held sets of beliefs about a group” (Biernat and Dovidio 2000, p. 108).

Stereotyped advertisements normally hold an advantage over counter-stereotyped advertisements, as consumers can use heuristics to process and remember the information better (Grier and McGill 2000). Although they have been drastically toned down compared to previous decades (e.g. ”Frito Bandito” that was the cartoon mascot for Fritos until 1971), advertisements are still characterized by stereotypes (e.g. Uncle Ben’s rice) (Bennett, Hill, and Oleksiuk 2013). Such portrayals have received vast criticism of nurturing age-old biases towards minorities (Bailey 2006). In spite of this, stereotyped portrayals in media communications still persevere.

With the rise of the digital world, marketing campaigns have become morepersonalized - targeting consumers individually by displaying advertisements specifically meant for each person (Mulhern 2009). However, this ”perfect personalization” also brings with it less attractive consequences, as it has been proven to discriminate consumers by displaying varying advertisements to different types of groups (Amit Datta, Tschantz, and Anupam Datta 2015). As a result, this new-age, algorithm- based marketing contributes to the continuous distortion of the real opportunities people face.

In a study conducted at Harvard, it was revealed how search advertisements on Google differed varying on whether the names typed in were typically ”Black” or ”White” (Sweeney 2013). More specifically, advertisements containing the word ”arrest” were shown for more than 80 percent of

”Black” name searches, while fewer than 30 percent for ”White”. Although algorithmic systems can theoretically be designed to help prevent any bias, they are built by humans and rely on behavioral data - which means they in reality mimic the biases present in the real world (Chander

(19)

2016).

Measuring Racial Discrimination

Although racial discrimination is still evident in most societies, the way people discriminate has significantly changed over the years. Racism was originally based on a purely biological foundation, however modern researchers agree this is no longer the case. Instead, ”modern racism” emphasizes the differences between ethnic groups in their languages, cultures and norms (Taguieff 1988).

Today, people thus mainly discriminate based on ethnic belonging (Gilroy 1991).

Moreover, the act of discriminating has evolved from overt and explicit to more covert and indirect discriminatory behavior (Pager and Shepherd 2008). For example, during the era of racial segregation in the U.S., minorities were overtly excluded from retail settings (”no Blacks allowed”) (Harris, Henderson, and Williams 1995). Today, many Blacks still face an underlying hostility in stores, for instance by being watched for shoplifting by salespeople while shopping in a store. Often disguised in a sophisticated manner that makes the effect difficult to isolate, racial discrimination is a challenging phenomenon to measure (Pager and Shepherd 2008).

A common way of studying discrimination is by investigating discrepancies in outcomes between ethnic groups (Pager and Shepherd 2008). This method involves examining the potential result of discrimination on unequal distribution of social and economic resources. However, a shortcoming of this approach is the inability to draw conclusions regarding causal relationships. One can only identify certain patterns, yet not fully conclude on discrimination being the sole factor explaining the outcomes. Other similar methods involve more detailed and systematic studies of individual cases, for example firms, in order to identify key factors that can better explain discrimination (Castilla 2008). In contrast, experimental approaches grant researchers the possibility of directly measuring causal effects by carefully adjusting the manipulations provided to respondents (Pager and Shepherd 2008). While laboratory experiments offers the strongest evidence of causal relationships, field experiments are also sometimes conducted in order to grasp the phenomenon in real-world contexts.

(20)

Within the fields of sociology and psychology, researchers studying biases have mainly utilized large-scale surveys focused on measuring people’s explicit attitudes; views and opinions that we are consciously aware of (Pager and Shepherd 2008). A typical weakness for this method is that people tend to answer in socially desirable ways in order to present themselves in a favorable manner (M. F. King and Bruner 2000). Social desirability bias, as it is called, may potentially compromise the validity of a study (Malhotra 1988). As a response, researchers have started to use measures meant to capture implicit attitudes - evaluations that operate subconsciously, but still influence cognition, affect and conduct (Greenwald and Banaji 1995). Therefore, implicit attitudes are considered more reliable reflections of actual biases and behavior (Dovidio et al. 1997). Rooth (2007) was the first ever to include the implicit association test (IAT) in his study of discrimination in the Swedish job market. The results showed a significant negative correlation between implicit attitudes and the callback rate for an interview for applicants with Arab/Muslim sounding names.

2.2.3 Racial Discrimination in The Sharing Economy

As mentioned in the introduction, the sharing economy has been considered as a potential antidote against racial discrimination in transactions (Leong 2015). The main reason seems simple: it removes the signals of race one would normally encounter when doing business in conventional marketplaces. As the parties are situated in different locations, as opposed to face to face in traditional marketplace-settings, it is presumably difficult for a business to discriminate against someone on the basis of race. Consequently, such online marketplaces have the potential to lessen discrimination by enabling more arms-length transactions (Edelman and Luca 2014). As described by Leong (2015, p. 2161): ”the idea is that the Internet — by filtering out racial signifiers — will eliminate the possibility of discrimination arising from overt or unconscious racism”. The internet can thus be argued to eliminate circumstances of discrimination, by acting as a neutral platform providing only the minimum information required to complete a transaction.

However, this argument assumes that sharing economy platforms do not present information that could potentially lead to any biases. In reality, this is not the case.

(21)

By allowing members to present personal information about themselves, the main intention of platforms such as Airbnb is mainly to build trust and facilitate transactions (Edelman and Luca 2014). When offered a rental request, Airbnb hosts are provided with the potential guest’s name, in most cases a picture, and other information that could possibly be of relevance (Todisco 2014). To increase the probability of being accepted, Airbnb encourages guests to share personal information with the hosts. Yet, these features can also bring unintended consequences by providing hosts with social cues as to who this person is. This is particularly the case when it comes to a trait as apparent as race (Leong 2015; Todisco 2014). What’s more, Leong (2015, p. 2162) argues that

”in addition to these instances of one-off discrimination, sharing economy businesses also employ rating systems that risk expression of implicit bias and even magnify its effects. Rating systems therefore instantiate the same inequality long present in the old economy”. As a result, online sharing economy platforms have been argued to do the opposite of providing a level field for all consumers (Leong 2015).

Although the right to full and equal access to any place of public accommodation has been strictly enforced in many countries, housing platforms within the sharing economy are not regulated under these laws (Leong 2015). While antidiscrimination laws ban the landlord of a large apartment building from discriminating based on race, it is argued that such laws do not reach many of the smaller landlords of the sharing economy (Todisco 2014). Concepts such as Airbnb are thus effortlessly circumventing this issue, by providing the individual landlords with the privilege of selecting their guests.

As mentioned in the introduction, in 2015, a Black man named Gregory Selden experienced the negative consequences of hosts being able to hand-pick their guests. In an attempt to achieve justice, he sued Airbnb for racial discrimination (Vara 2017). Even though Selden’s lawsuit was unsuccessful in holding Airbnb legally accountable, it quickly inspired many others to spread their own similar experiences throughout the internet, putting pressure on Airbnb to start fighting this issue. As a response, the company introduced its own nondiscrimination policy, which emphasizes the importance of acting based on the values inclusion and respect (Airbnb 2017b). However, the true effect of the company’s code of ethics remains questionable.

(22)

As current research depicts, Airbnb has been associated with facilitating both unintentional, as well as intentional, unregulated racial discrimination (Todisco 2014). As mentioned, an experiment conducted by Edelman, Luca, and Svirsky (2017) revealed that applications on Airbnb from guests with typically African-American names were 16 percent less likely to be accepted, compared to identical guests with typically White names. Another study, conducted by Edelman and Luca (2014), revealed that nonblack hosts were able to charge approximately 12 percent more than Black hosts, when keeping location, quality and rental characteristics constant. In addition, Black hosts received a larger price penalty for having a poor location, as compared to nonblack hosts.

Based on these studies, one can question whether the sharing economy is as inclusive as expected, or whether the saying “what’s mine is yours” (Botsman and Rogers 2010) paradoxically only holds true for certain groups of privileged people. As research depicts, racial discrimination is still apparent in many types of markets - including digital ones. This triggers the question; why is it that, even in such new and future-oriented economies, consumers are still utilizing age-old biases to make decisions? Put from a more general perspective: why do we discriminate?

2.2.4 Psychological Processes Behind Racial Discrimination

A central theory of social psychology that can help explain why discrimination still exists, is social identity theory (Tajfel and Turner 1979). This theory provides a psychological perspective for studying country-of-origin effects, by connecting in-group bias with stereotypes of in-groups and out-groups (Verlegh 1999).

Social Identity Theory

One way to study the rationale behind people’s behaviors, is through the assumption that individuals do what they do because of who they believe they are – their identity (Korte 2007).

The core concept of the self “embodies personal history, relates the individual to social situations, shapes cognition, and anchors a range of goals, motives, and needs” (Turner and Onorato 1999, pp. 15-16). Identity is thus a relational and self-referential cognitive construct of the self, that answers the question “Who am I?”. Moreover, individuals encompass multiple selves or identities

(23)

(Fiske and S. E. Taylor 1991; Hogg, Terry, and K. M. White 1995). For example, a woman can characterize herself as both a mother, wife and business person.

Surrounding the core concept of the self are the more peripheral concepts, which allow an individual to adapt to various social situations through group identities (Korte 2007). These various roles are classified as an individual’s social identities. The concept of social identity is defined as ‘‘that part of an individual’s self concept which derives from his knowledge of his membership of a group (or groups) together with the value and emotional significance attached to the membership” (Tajfel 1978, p. 63). Social groups can exist at multiple levels, including societal, cultural, industrial, organizational, functional and professional) (K. M. White, Hogg, and Terry 2002). Furthermore, the strength of a specific identity is related to the individual, the group, and the context (Ashforth and Mael 1989; Turner, Hogg, et al. 1987).

According to social identity theory, individuals tend to engage in intergroup differentiation – by attempting to maximize differences between the in-group (the group to which one psychologically belongs) and the out-group (the psychologically relevant opposition group) (Tajfel and Turner 1986). This mechanism is driven by the primary motivations of desire for certainty and positive self-evaluation (Abrams and Hogg 1990; Hogg and Grieve 1999). A crucial aspect of social identity theory is that social classifications are not justified by any formal group membership, rather self- perceived membership in a specific group (Greene 2004). Intergroup differentiation occurs in two primary ways: in-group-favoritism – by exaggerating and enhancing the favorable qualities of its members, and out-group derogation – by exaggerating the negative characteristics of relevant out- groups (Brewer and Brown 1998). These two processes do not necessarily need to co-occur. Either way, the result of each process is an enhanced distinction between groups, through establishing the superiority and status of the in-group.

Intergroup Differentiation and Race

The powerful influence of social identity can be witnessed in individuals displaying intentions and conducts that would originally conflict with their personal identities (Ashforth and Mael 1989). In particular, race can serve as a visually salient and stable cue for group categorization, by providing well-established associations as a basis for in-group favoritism and/or out-group derogation (Brewer

(24)

1979; Greenland and Brown 1999; Cosmides, Tooby, and Kurzban 2003). A particularly startling aspect of intergroup differentiation is that people have been shown to respond in an automatic and uncontrollable manner to even unconscious exposure to outgroups (e.g. Devine 1989; Dovidio et al. 1997; Greenwald and Banaji 1995). This is especially observable when it comes to race:

ordinary cognitive processes can trigger automatic racial biases, even when an individual endorses egalitarian beliefs (Ito and Urland 2003; Gaertner and Dovidio 1986). Automatic biases therefore act as an essentially unescapable part of intergroup perception (Van Bavel and W. A. Cunningham 2009).

Racial biases have been proven exceedingly prominent and problematic to defeat (Park and Roth- bart 1982). As a result, it is unsurprising that racial bias still persists in our world today (Blank, Dabady, and Citro 2004). Put eloquently by The National Research Council: “People’s intentions may be good, but their racially biased cognitive categories and associations may persist. The result is a modern, subtle form of prejudice that goes under-ground so as not to conflict with antiracist norms while it continues to shape people’s cognitive, affective, and behavioral responses. Subtle forms of racism are indirect, automatic, ambiguous, and ambivalent” (NRC 2004, p. 59).

Consumer Outcomes of Intergroup Differentiation

Stereotypes often arise as a result of out-group derogation, leading to prejudice and conflict as critical outcomes of social identity and self-categorization (Tajfel 1982; Turner, Hogg, et al. 1987).

Moreover, exposure to stereotypes and prejudices over long periods of time generates even deeper and more profound associations (A. W. Staats and C. K. Staats 1958). Individuals that hold strong racial biases are more prone to engage in discriminatory behavior (Greenwald, Poehlman, et al.

2009), including explicit discrimination (Rudman and Ashmore 2007). Yet, implicit, automatic reactions to outgroup features can also lead to discriminatory behavior (Bargh and Chartrand 1999), fear and anxiety (Frantz et al. 2004; Phelps et al. 2000) and negative stereotypic associations (Olson and Russell H. Fazio 2003).

Whether it is explicit or hidden, racial bias has been proven to have a pronounced influence on consumer behavior. In-group members have empirically been proved as more likely to receive positive valuation from other in-group members (Brewer 1979) as well as achieve more affect and

(25)

trust than out-group members (Kramer and Brewer 1984). In addition, individuals belonging to an in-group also enjoy more cooperative behavior from other in-group members (Schopler and Insko 1992: Dawes, Van De Kragt, and Orbell 1988). Within the field of economic decision making, the concept of social identity is noteworthy to investigate (Akerlof and Kranton 2000). Research has shown that in-group members impulsively allocate resources to their fellow group-members, consequently harming the out-group (Brewer and Brown 1998; Frey and Bohnet 1997).

In conclusion, racial biases undoubtedly have a profound impact on how we intentionally – and unintentionally – choose to interact with others. However, in order to fully grasp the hidden causes of contemporary racial discrimination, it is essential to study which specific internal states cause this behavior (Pager and Shepherd 2008). This can be done by moving from ”motives” to

”mechanisms” (Reskin 2005). Therefore, in order to learn more about underlying factors that may shape racial biases, we now shift the focus towards more specific interpersonal concepts and their relevance in the sharing economy sphere.

2.2.5 Theories of Congruence

A relevant aspect within the area of consumer psychology that can aid in explaining racial discrimination is congruence. Individuals generally seek to maintain cognitive consistency (Abelson et al. 1968), and act in ways that preserve and enhance their self-concept (Graeff 1996). This can be done by associating with people and objects we feel are consistent with our own perceptions of ourselves (Britt 1966; A. Aron and E. N. Aron 1996). In the following section, two key psychological concepts that can be linked to discrimination within consumer behavior will be presented:

self-other overlap and self-brand connection.

Self-other Overlap

A key concept within consumer-to-consumer interactions is self-other overlap, which sheds light on how we distinguish between the self and others. Research in social psychology has proposed that people sometimes psychologically include others in the self (A. Aron and E. N. Aron 1986; A.

Aron, E. N. Aron, and Smollan 1992; Myers and Hodges 2012). As representative of a psychological

(26)

construct, self-other overlap refers to a sense of oneness and ”shared or interconnected identities with others” (A. Aron, E. N. Aron, Tudor, et al. 1991). Put differently, it refers to overlapping mental representations between the self and others. The psychological construct of self-other overlap is more or less directly accessible to respondents (Myers and Hodges 2012) meaning that we almost immediately can establish our interpersonal closeness to any other person. Self-other overlap can form with any partner, regardless of affinity (Myers and Hodges 2012). In addition, it varies across social contexts, and adapts easily to changing input from the social environment.

Self-other overlap develops and endures for several reasons, one key reason being people’s desire to expand oneself (A. Aron and E. N. Aron 1986). As people perceive another person as part of the self, allocation of resources becomes mutual, actor-observer perspective differences are lessened, and the other’s characteristics become one’s own. As a result, this self-other overlap enhances a person’s self-efficacy, intrinsic motivation, and self-actualization (A. Aron and E. N. Aron 1996;

A. Aron, Melinat, et al. 1997; A. Aron, Norman, and E. N. Aron 1998; A. Aron, Norman, and E. N. Aron 2001).

A strong sense of self-other overlap can influence how information about the other person is processed. By contributing to a lessened self-other distinction at the cognitive level, it opens up for a more complex understanding of others (A. Aron and E. N. Aron 1986). Galinsky and Moskowitz (2000) found that self-other overlap had an effect on more positive evaluations of the target person and less stereotypical judgments of that person’s group. Moreover, research has proved how strong self-other overlap leads to greater valuation of and commitment to the other person, as well as relationship satisfaction (A. Aron, E. N. Aron, and Smollan 1992; Agnew et al.

1998; A. Aron and Fraley 1999).

In contrast, a weak feeling of self-other overlap - a low sense of”oneness” - does the exact opposite.

Based on a similar assumption central to social identity theory, we use certain clues to decide how we will ultimately behave towards others (Tajfel and Turner 1979 as cited in Schubert and Otten 2002). As the merging of self and in-group increases, so does the favoritism toward the in-group (Turner, Hogg, et al. 1987). With a low self-outgroup overlap, stereotypes will be activated when evaluating the person (Galinsky and Moskowitz 2000), consequently influencing how we act towards him or her. When it comes to ethnicity, we therefore tend to favor the in-group and thus

(27)

disadvantaging the out-group (Cadinu and Rothbart 1996; E. R. Smith and Henry 1996).

Self-brand Connection

As mentioned previously, the concept of “self” is something within every human that we use to describe ourselves, and is proposed to have an important influence on the brands we consume (Sameeni and Qadeer 2015). Self-brand connections are defined as“the degree to which consumers have incorporated the brand into their self-concept”(Escalas and Bettman 2003, p. 340). People use products and brands to create and represent self-images, reinforcing and expressing self-identity, and differentiating oneself (Belk 1988; Richins 1994). In this process, a link bridges the brand and the self. Thus, self-brand connections are proposed to capture an important part of consumers’

construction of self (Escalas and Bettman 2003).

Self-brand connections can have a favorable effect on brand attitudes and behavioral intentions (Escalas 2004). Consumers respond more favorably to brands connected to their sense of self, and help them achieve their self-identity goals. There should also be a positive relation between self-brand connections and consumers’ likelihood of trial, purchase, higher willingness to pay, or all of these, as consumers with self-brand connections behave more consistently with regard to the brand.

Research on reference groups related to consumer self-brand connections shows that consumers have a stronger self-brand connection to brands consistent with an in-group (Escalas and Bettman 2005). As self-brand connections are important in our construction of our self, we wish to be associated with the kind of people consuming the brands we choose, while also avoiding brands congruent with out-groups. Doing so may harm the consumers’ self-image (R. E. Kleine, S. S.

Kleine, and Kernan 1993); Wooten and Reed 2004). Associations about reference groups become associated with brands those groups are perceived to use and vice versa (Escalas and Bettman 2003).

Social identity theory can be used to explain reference group influences on self-brand connections.

People strive for positive distinctiveness from out-groups (K. White and Dahl 2006). As ethnicity can serve as a cue for group categorization, consumers may avoid brands associated with different

(28)

ethnic groups. K. White and Dahl (2006) found that dissociative reference groups, e.g. a group a person wants to avoid, have a greater impact on consumers’ negative self-brand connections, product evaluations, and choices than do products associated with out-groups more generally.

Studies also find that service brands, which is the category Airbnb falls under, have the potential to engage consumers at the self-concept level (Dwivedi 2014). Consequently, the concepts of self- brand connection and reference groups may be applied to the sharing economy environment.

2.2.6 Theories of Trust and Risk

In addition to the theories of congruence, we have identified two other distinct concepts that can be applied to both the field of racial discrimination and online transactions in the sharing economy.

These concepts are trust and perceived risk, and are both proved to be important in the area of consumer behavior research - in particular when it comes to the digital sphere (Kim, Ferrin, and Rao 2008; Ratnasingham 1998; Harridge-March 2006).

Trust

The role of trust in online economic exchange has been studied extensively, as this geographically dispersed, non-face-to-face transaction creates a significant risk of opportunism (Bapna et al. 2017;

Lee 2015). As sharing economy platforms are working peer-to-peer without middlemen, trust is one of the fundamental principles for collaborative consumption to work (Botsman and Rogers 2010). Trust can be defined as“the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor” (Mayer, Davis, and Schoorman 1995, p. 712). Further, Coleman (1995) defined trust as “a willingness to commit to a collaborative effort before you know how the other person will behave”. Both definitions emphasize how trust is about the willingness to take risks, and the willingness to make oneself vulnerable.

Trustworthiness relates to the trusting attributes of the trustee (a party to be trusted) (Mayer, Davis, and Schoorman 1995). According to social identity theory, the more a person differs from oneself, the less likely it is that we will trust this person (Tajfel 1982; Messick and Mackie 1989,

(29)

Brewer and Pierce 2005). Ethnicity is one such basis of difference that may lead to differences in trustworthiness. Zucker (1986) identified characteristics-based trust as being formed within a group on the basis of factors such as ethnicity. Cross-national surveys show that trust is lower in heterogeneous countries (Knack and Keffer 1997). Effects are also found in local communities, as diversity reduces generalized faith in others (Alesina and La Ferrara 2000). In the Norwegian society, the overall level of trust is relatively high, meaning that people generally trust the society’s institutions as well as ”most people” (Ortiz-Ospina and Roser 2016). In addition, the level of trust tends to increase for highly educated people.

Trust is a multidimensional concept that can be studied from the viewpoints of social psychology, sociology, economics, and marketing (Doney and Cannon 1997). For online platforms, trust has shown to be the single most important factor that determines consumers’ transactions with a vendor (Gefen 2000). S. W. Wang, Ngamsiriudom, and Hsieh (2015) found a positively significant relationship between trust andbehavioral intention. Trust in the online platform may also directly influence theattitude towards internet shopping (Jarvenpaa, Tractinsky, and Vitale 2000).

Benevolence is an important attribute of trust, and can be explained as the extent to which a trustee is believed to want to do good to the trustor, aside from an egocentric profit motive (Mayer, Davis, and Schoorman 1995). Benevolence can for example be how a mentor (trustee) wants to help a trustor without it being required, or having any extrinsic motivation. Benevolence suggests that the trustee has some specific attachment to the trustor, and high benevolence in this relationship would be inversely related to the motivation to lie (Hovland, Janis, and Kelley 1953). Ethnic diversity is argued to be negatively related to benevolence (Parboteeah, Seriki, and Hoegl 2014). Diversity encourages caring for those within the same ethnic group (in-group), at the expense of the out-group. The perceived benevolence from a out-group member can thus be reduced.

Hu et al. (2016) have shown how benevolence of peer members in an online shopping community positively impacts the shoppers’ utilitarian value. This perceived utilitarian value predict individuals’ consumption intentions. Further, Lu, Zhao, and B. Wang (2010) found how trust in members’

benevolence stimulated the purchase intention in consumer-to-consumer e-commerce settings. In conclusion, trust, including the dimension of benevolence, seems to play an important role in both

(30)

the contexts of intergroup differentiation and digital platforms.

Perceived Risk

Many rational-choice economic models fail to explain how individuals actually calculate risk, as we are not always able to make informed predictions about the likelihood of future events. Social and cognitive psychological experiments show that distracting information bias our predictions, and that we are likely to look to social or contextual factors for additional information (Tversky and Kahneman 1975; Douglas and Wildavsky 1982; Short 1984). The theory of consumer perceived risk was first introduced by Bauer (1960). Bauer defined perceived risk as”the unpredictable results that consumers perceive when they engage in purchasing behavior”.

Airbnb’s business model consists of online transactions, where the booking and payment is placed without any physical examination of the rental space, and no face-to-face contact with the seller beforehand. Online transactions involve more uncertainty than the traditional brick-and-mortar purchases, increasing the level of consumer perceived risk (Brynjolfsson and M. D. Smith 2000).

In addition, evidence suggests that services may be perceived as particularly risky, due to their fundamental nature (Guseman 1981; Murray and Schlacter 1990). As a result, consumers will likely demand increased information for predominantly service-type products (Deshpande and Krishnan 1977).

Race is one influence that particularly contributes to individuals’ perception of risk (Quillian and Pager 2010). This effect can be explained by the social amplification of risk, which refers to amplifying the relevance of certain factors while downplaying others (Kasperson 1988). The amplification can occur where the message is sent, such as extensive media coverage, or by the receiver, by adding social values and meaning to the message (Quillian and Pager 2010). A report from the Norwegian directory of Integration and Diversity from 2009 shows that 71 % of Norwegian media stories with immigrants or immigration as the main topic are considered to be problem-oriented (IMDi 2009). This negative media coverage is thus one factor that can amplify the perceived risk of engaging with immigrants.

(31)

S. M. Cunningham (1967) classified risk into six dimensions, which later has been modified by different scholars along with the emerge of online shopping. Luo et al. (2010) suggest seven dimensions of risk based on Cunningham’s work, which are applicable to the context of Airbnb;

functional, time, financial, privacy, security, psychological, and social risk. The dimensions of perceived risk have been identified as important attitudinal factors that influence adoption behavior (Luo et al. 2010; Laforet and X. Li 2005).

2.2.7 Individual Differences and Racial Bias

In addition to the abovementioned concepts, there are certain person-specific factors that may be influential in determining whether a person is prone to engage in racial discrimination. In contrast to the other concepts, these inherent aspects do not change according to the situation an individual finds himself or herself in. Hence, these are generally stable traits or attitudes that can affect how individuals perceive the world, and consequently act in it.

Political Orientation

Traditionally, people’s social class has seemed to be the main determinant of their political orientation (Lipset and Rokkan 1967). However, in liberal countries (as for example Norway), personal values have shown to predict political orientation more strongly than sociodemographic variables over the last years (Piurko, Schwartz, and Davidov 2011). Individuals’ left-right political orientation can thus be said to express their personal values. For left-orientated parties, equality and concern for others has traditionally been the main motivation (Piurko, Schwartz, and Davidov 2011). Avoiding change and controlling threats typically motivate a right orientation. This is also reflected in the respective immigration policies of the parties, ranging from right to left. For far right-wing parties in Europe, ethnic nationalism is a common ideological feature (Golder 2016).

Ethnic nationalism is exclusionairy, and focuses on obtaining a monocultural state by limiting immigration. Although the common consensus does not support the extreme ideologies of far-right parties, several right-oriented parties in Norway have lately started to move towards a stricter immigration policy (Simonnes 2013). In contrast, left-orientated parties generally have a more liberal immigration policy.

(32)

In the U.S., research on political orientation and race has portrayed how White conservatives (generally right-wing oriented) overall still maintain prejudiced beliefs and attitudes towards Blacks, although they typically conceal such beliefs in most contemporary public settings (Nail, Harton, and Barnes 2008). However, when they are not aware of their behavior being monitored, they often reveal their prejudice through discriminatory actions (Nail, Harton, and Decker 2003). In addition, conservatives have been shown to feel lower obligation to conceal their implicit racial bias and consequently avoid discriminating (Redford and Ratliff 2016). Compared to conservatives, liberals (generally left-wing oriented) typically possess more subtle, indirect expressions of prejudice, and even though they tend to hold more egalitarian attitudes like fairness and equality, these negative, race-based feelings do surface from time to time (Hing, W. Li, and Zanna 2002). However, in situations where race and norms of fairness are both salient, liberals have been proven to actually show favoritism to Blacks - referred to as thereverse discrimination effect (Nail, Harton, and Barnes 2008). Political orientation has thus been included in several studies related to racial discrimination, often revealing distinct differences in outcomes depending on people’s political views.

Intergroup Threat

As emphasized previously, research within psychology has shown how prejudice just may be an inevitable aspect of human life (Allport 1954). In this context, an important concept to investigate is intergroup threat, which occurs when “one group’s actions, beliefs or characteristics challenge the goal attainment or well-being of another group” (Riek, Mania, and Gaertner 2006, p. 336). Put in other words, intergroup threat involves what might frighten one group (or individual) about another group (or individual), and the resulting perceptions and actions stemming from those fears. The term is person-specific, meaning that different individuals can feel different degrees of intergroup threats. The concept can be utilized when investigating intergroup prejudice, negative outgroup attitudes and their subsequent injustices (Riek, Mania, and Gaertner 2006; Sabbagh and Schmitt 2016).

Stephan and Renfro (2002) unveiled how intergroup threat can predict attitudes toward racial outgroups in both White and Black samples. Moreover, the concept has also been proven to explain attitudes toward various immigrant groups (Bizman and Yinon 2001; Cur¸seu, Stoop, and

(33)

Schalk 2007; Rohmann, Florack, and Piontkowski 2006, Stephan, Renfro, et al. 2005, Stephan, Ybarra, and Bachman 1999, Stephan, Ybarra, Martnez, et al. 1998). In general, as stated by Riek, Mania, and Gaertner (2006, p. 345): “as people perceive more intergroup competition, more value violations, higher levels of intergroup anxiety, more group esteem threats, and endorse more negative stereotypes, negative attitudes toward outgroups increase”. Furthermore, when members of an in-group feel endangered by the respective out-group, they tend to behave negatively towards the out-group (Hart et al. 2000, Phelps et al. 2000, Bargh and Chartrand 2000). This will either be by subtly ignoring them, or harming them by more explicit means, by systematically discriminating members of the out-group.

2.2.8 Eradicating Racial Bias

As emphasized throughout this literature review, our thoughts and behaviors are easily influenced by the social cues we encounter in our every day lives. Acting as simplifying heuristics, we use social categories to guide our perceptions and evaluations of others (Turner, Hogg, et al. 1987;

Turner, Oakes, et al. 1994). As mentioned previously, one particularly visually noticeable social cue is race. Research has shown that racial biases are immediately activated just by seeing a person’s face or name (Russel H. Fazio et al. 1995). On the other hand, we have multiple social identities, and the category that best fits a specific social context becomes salient in that particular case (Oakes and Turner 1990). As a result, biases can potentially be eliminated by manipulating the focus towards a more inclusive superordinate identity (Gaertner and Dovidio 2000).

This phenomenon is highlighted in the Common Ingroup Identity Model (CIIM) (Gaertner and Dovidio 2000). According to this model, one can reshape group boundaries through a process of recategorization. Cross-cutting social categories can become salient through the introduction of a shared bond between two groups (in-group and out-group) (Riek, Mania, and Gaertner 2006). As a result, intergroup bias will diminish, and former out-group members may suddenly receive the same preferential treatment as original in-group members. In a study conducted by Van Bavel and W. A. Cunningham (2009), the researchers investigated whether self-categorization with a new mixed-race group would defeat automatic racial bias. As predicted, Black in-group members

(34)

received more positive automatic evaluations than Black out-group members. This shows that just by receiving a hint that someone is more ”like us”, prejudice may melt away and leave more favorable and embracing outcomes.

2.3 Summing Up

In conclusion, the issue of racial discrimination seems to be an almost inevitable aspect of ethnically diverse societies. Although it has evolved from overt and explicit into covert and implicit (Pager and Shepherd 2008), it is still present in areas such as traditional and digital marketplaces, as well as in marketing campaigns. Consequently, racial discrimination continues to systematically affect the allocation of resources in most societies. As revealed by Edelman, Luca, and Svirsky (2017), evidence of racial discrimination was even found within the disruptive digital market called the sharing economy. Characterized by its innovative way of connecting consumers directly to each other, the sharing economy has proved to be a free ground for people to make their decisions based on their own biases and preferences (Leong 2015; Todisco 2014).

Although racial discrimination is a complex phenomenon (Pager and Shepherd 2008), social identity theory can help us understand why we develop biases and discriminate towards certain others.

With the ultimate goal of cultivating our own identity - our core concept of ourselves (Turner and Onorato 1999), we engage in intergroup differentiation in order to satisfy our desire for certainty and positive self-evaluation (Abrams and Hogg 1990; Hogg and Grieve 1999). However, in order to fully understand the hidden causes of contemporary racial discrimination in marketplaces, we must shift the focus frommotives towardsmechanisms (Reskin 2005). This will be discussed further in the following section regarding our contribution to the literature.

2.4 Our Contribution to the Literature

The issue of racial discrimination has been studied extensively within a broad range of social and economic areas, by researchers from academic fields such as law, economics, sociology and consumer

(35)

psychology. It is therefore a well-documented phenomenon in most contexts of society. Moreover, the influx of disruptive digital markets has created a new economy based on consumers sharing their resources directly with each other. Empirically proven by Edelman, Luca, and Svirsky (2017), it has been revealed that the sharing economy platform called Airbnb facilitates unregulated racial discrimination (Todisco 2014).

Modern researchers agree that racism is no longer based explicitly on race. Instead, it emphasizes the differences between ethnic groups in their languages, cultures and norms (Taguieff 1988). Espe- cially during the last year, the world has witnessed an increase in xenophobia in Western countries, involving fear or hatred of people from different cultures (Roth 2017). This has particularly been directed towards people from predominantly Muslim countries, such as Arabs from Middle Eastern nations.

In this thesis, we will investigate the issue of ethnic bias further, by drawing inspiration from the experiment carried out by Edelman, Luca, and Svirsky (2017) in the sharing economy. As their study was conducted in the U.S. through a field experiment, we will contribute to the literature by examining the issue on a new population (in Norway), and instead employing an online experiment based on a survey. This way, we can examine the causal relationship between ethnicity and discrimination in the sharing economy in a more controlled manner. Moreover, we will explore the potential effect of introducing a common identity in order to eliminate bias.

Furthermore, this research design will enable us to investigate the mechanisms by which the effect of ethnic discrimination operates, as well as establishing its boundary conditions. More specifically, we will investigate mediating effects related to the theoretical concepts self-other overlap, self-brand connection, trustworthiness and perceived risk. In addition, the person-specific traits political orientation and intergroup threat will be included as potential moderators, to learn more about which types of people discriminate. Answering these questions of how and when may provide us with a deeper understanding of the phenomenon of racial discrimination (Hayes 2013), sparking useful insights for reducing the problem in the future.

(36)

Chapter 3: Research Model and Hypotheses

In this chapter, we will present the proposed research model and its corresponding hypotheses.

This will be used for answering our research questions, presented in the main purpose of this study.

3.1 Proposed Model

Figure 3.1: Proposed Model of Effects With Hypotheses

Figure 3.1 is a visual representation of our hypotheses, and summarizes the relationships we will investigate. It shows the hypothesized effects from the independent variable on the dependent variables, both directly and indirectly through four mediators. We propose that host ethnicity will influence both the outcomes attitudes and behavioral intentions. We consider this effect to be direct, as well as mediated through self-other overlap, self-object connection, trustworthiness

(37)

and perceived risk. In addition, we believe these direct and indirect processes are moderated by political orientation and intergroup threat.

3.2 Hypotheses

In order to answer our research questions, we propose nine hypotheses. Hypotheses H1-H3 aim to test the direct effects of the host’s ethnicity on attitudes and behavioral intentions, which will be referred to as consumer outcomes. Moreover, hypotheses H5-H8 are concerning how the relationship between host ethnicity and consumer outcomes is mediated. Finally, hypotheses H4 and H9 involve the boundary conditions for both the direct and mediated effects.

3.2.1 Consumer Outcomes

As specified, consumer outcomes refer to our dependent variables, namely the attitudes and behavioral intentions regarding two Airbnb service scenarios. Previous research has established how discrimination based on race is apparent in a vast array of market transactions (Hanson, Hawley, and A. Taylor 2011; Bye et al. 2014; Nunley, Owens, and Howard 2015). Within the field of psychology, out-group members (such as those of another race) have empirically been proved to receive less favorable outcomes, compared to in-group members (e.g. Brewer and Brown 1998;

Frey and Bohnet 1997). Edelman, Luca, and Svirsky (2017) unveiled how this phenomenon also exists in the sharing economy. As seen in the literature review, people now mainly discriminate based on ethnic belonging (Gilroy 1991). As we will discuss later on in the Methodology chapter, we chose to use the Arab ethnicity due to its stigmatization in the Norwegian society.

In order to minimize the effect of social desirability bias (M. F. King and Bruner 2000; Malhotra 1988), we will investigate the relationship between host ethnicity and several types of attitudes and behavioral intentions. We thus propose the following hypothesis:

Digital discrimination in the sharing economy: what’s mine is not yours : evidence from an online experiment

Norwegian School of Economics